Small-World Minds Create the Illusion of Low-Dimensional Social Cognition Under Constrained Stimuli

A long-standing tradition suggests that social inferences are summarized by a few dimensions (e.g., warmth, competence). However, such findings relied on constrained stimuli (e.g., group labels, static faces). Studies using naturalistic designs instead reveal high dimensionality. We propose a new perspective that reconciles these views: mental representations of social inferences are structured as a growing network with small-world properties. In this structure, nodes (social inferences) are connected with only a few steps from one another, so even limited activation—elicited by constrained stimuli in only a small subset of nodes—propagates broadly across the network, producing covariation that mimics low-dimensional structures. However, these dimensions are not functional but emergent properties of network dynamics under constrained input. We demonstrated how this small-world-minds perspective accounts for classic phenomena—halo effect and psychological dimensions—as artifacts of stimulus limitations. We call for moving beyond low-dimensional paradigms to understand the complexity of social cognition.